Written and maintained by CASRAI Editorial Board
Last updated
The Social Vulnerability Index (SVI) is a CDC/ATSDR measure that ranks every US census tract by how vulnerable its residents are likely to be when a disaster or public health emergency strikes. Researchers, emergency planners and health analysts use it to compare places. This guide shows how the score is built, how to compute a rank-and-sum index of the same kind yourself, and how to read and report the result without over-claiming.
Definition in one paragraph
CDC/ATSDR describe the SVI as a ranking of each census tract’s relative social vulnerability, using 16 variables from the American Community Survey (ACS) grouped into four themes. The 2022 release uses ACS 2018-2022 five-year estimates. Each tract receives a percentile rank for every variable, for every theme and for all variables together. Percentile values run from 0 to 1, and higher values mean greater social vulnerability. The index is relative: it says where a tract sits compared with other tracts, not how much harm a tract will suffer. The original index (Flanagan et al., 2011) used 15 census variables; the first database was released in 2011.
CDC SVI themes and variables
| Theme | Variables (2022 release) |
|---|---|
| Socioeconomic status | Unemployed (age 16+); no high school diploma (age 25+); persons below 150% of the poverty level; housing cost burden (30% or more of income); no health insurance |
| Household characteristics | Age 65 and older; age 17 and younger; civilian population with a disability (age 5+); single-parent households; English language proficiency below “well” |
| Racial and ethnic minority status | One variable |
| Housing type and transportation | Housing in structures with 10 or more units; mobile homes; crowding (more people than rooms); no vehicle available; group quarters |
Adjunct variables are published alongside the data but are not used in the rankings.
How is the Social Vulnerability Index calculated?
The calculation is a rank-and-sum chain: rank each variable, add the ranks within a theme, rank the theme sums, then repeat once for the four themes together. Because every step is a ranking, no variable is measured in its own units once ranking begins, so a percentage and a count can be combined.
Social vulnerability index formula
CDC/ATSDR name the output columns as follows. Each symbol is one column in the downloadable database.
- EPL_: the percentile rank of one estimated percentage variable across all tracts in the database (16 of these).
- SPL_: the sum of the EPL_ values for the variables in one theme.
- RPL_: the percentile rank of that theme sum (SPL_), giving one theme ranking per tract (RPL_THEME1 to RPL_THEME4).
- SPL_THEMES: the sum of the four theme sums. CDC/ATSDR note that summing the theme sums is the same as summing all individual variable rankings.
- RPL_THEMES: the percentile rank of SPL_THEMES. This is the overall SVI ranking.
- F_ flags: a variable is flagged (1) when the tract is in the top 10% (90th percentile) for that variable; theme flags count the variable flags in the theme; the overall flag counts all variable flags.
In short: RPL_THEMES = percentile rank of the sum, across the four themes, of the sums of EPL_ values.
Worked example with the arithmetic shown
This is an illustrative toy example with invented numbers, not CDC data. It uses four tracts and two variables, and the convention percentile = (rank – 1) / (N – 1) with N = 4, so ranks 1 to 4 become 0, 0.333, 0.667 and 1. That convention is used here to make the arithmetic visible; it is our simplification, so do not expect it to reproduce CDC values exactly.
| Tract | % below poverty line (x) | % no vehicle (y) | EPL x | EPL y | SPL = EPL x + EPL y | RPL |
|---|---|---|---|---|---|---|
| A | 10 | 5 | 0.000 | 0.333 | 0.333 | 0.000 |
| B | 25 | 3 | 0.667 | 0.000 | 0.667 | 0.333 |
| C | 40 | 20 | 1.000 | 1.000 | 2.000 | 1.000 |
| D | 18 | 12 | 0.333 | 0.667 | 1.000 | 0.667 |
Step 1: rank x from lowest to highest: A (10) = 1, D (18) = 2, B (25) = 3, C (40) = 4, giving EPL x of 0, 0.333, 0.667 and 1 for A, D, B and C. Step 2: rank y the same way: B (3) = 1, A (5) = 2, D (12) = 3, C (20) = 4, giving EPL y of 0, 0.333, 0.667 and 1 for B, A, D and C. Step 3: add: A = 0 + 0.333 = 0.333; B = 0.667 + 0 = 0.667; C = 1 + 1 = 2; D = 0.333 + 0.667 = 1. Step 4: rank the sums (A 1, B 2, D 3, C 4) and convert to percentiles to get the RPL column. Tract C is the most vulnerable on this toy index; tract A the least. With real data the same steps run across tens of thousands of tracts and 16 variables.
How to interpret SVI scores
| What you see | Reading guidance (from CDC/ATSDR documentation) |
|---|---|
| RPL_THEMES near 0 | Lower relative social vulnerability compared with the other tracts in the database used. |
| RPL_THEMES near 1 | Higher relative social vulnerability. It is a rank position, not a probability or a count of people affected. |
| Theme rankings (RPL_THEME1 to 4) | Show which dimension drives a tract’s overall position; the overall ranking can hide a tract that is extreme on one theme. |
| Flag count | The number of variables on which the tract is in the top 10%. |
| Value of -999 | The value is unavailable or could not be calculated; it is not used in later calculations, and total flags exclude it. |
| Tract with zero estimated population | Excluded from the ranking (857 tracts in the US 2022 database). |
The same tract can have different rankings in the national and state databases. The national database ranks all US tracts against each other; each state database ranks tracts only within that state. CDC/ATSDR’s documentation shows a Fulton County, Georgia example in which the 2018 overall ranking was 0.2658 in the state database and 0.5268 in the national one. Use the state database for within-state analysis.
Assumptions and common mistakes
- Treating the ranking as a score. RPL values say where a tract sits among tracts, not how vulnerable it is in absolute terms. A change from 0.60 to 0.70 means a change in rank position, not a 10-point rise in vulnerability.
- Mixing databases. Do not compare a state-database ranking with a national one, or rankings from releases that use different variable sets (15 variables in 2011, 16 in the 2022 release) or different ACS years, as if they measured the same thing.
- Equal weighting is a design choice. Every variable enters the sum with the same weight; the index does not claim that, for example, vehicle access matters as much as poverty in every hazard.
- Ecological inference. A tract rank describes an area, not the individuals living in it. See the ecological study design and the ecological fallacy guide.
- Ignoring excluded or -999 tracts. Report how many tracts were missing in your analysis area.
For other composite area measures, see the Index of Multiple Deprivation, and for how percentile ranks work in general, see percentile and quantile and how to calculate a percentile from a rank.
How to report it (APA example)
State the release year, the database (national or state), the ranking you used, and the ACS years. Example sentence: “Census tracts were classified by the overall Social Vulnerability Index ranking (RPL_THEMES) from the CDC/ATSDR 2022 national database, which is based on ACS 2018-2022 estimates; tracts in the top quartile (RPL_THEMES of 0.75 or higher) were treated as high vulnerability.” The quartile cut-off is an analyst’s choice and should be stated as such. Reference entries in APA 7th style:
Agency for Toxic Substances and Disease Registry. (2024). CDC/ATSDR Social Vulnerability Index 2022 documentation. U.S. Department of Health and Human Services. https://www.atsdr.cdc.gov/place-health/media/pdfs/2024/10/SVI2022Documentation.pdf
Flanagan, B. E., Gregory, E. W., Hallisey, E. J., Heitgerd, J. L., & Lewis, B. (2011). A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management, 8(1), Article 3. https://doi.org/10.2202/1547-7355.1792
How to calculate it in Excel, R or Python
These snippets reproduce the toy example above (rank, convert to a 0 to 1 percentile, sum, rank again). They are a teaching recipe for the rank-and-sum logic, not a replacement for the CDC/ATSDR data files, which already contain the published columns.
Excel
EPL x (cell D2): =PERCENTRANK.INC($B$2:$B$5, B2, 15)
EPL y (cell E2): =PERCENTRANK.INC($C$2:$C$5, C2, 15)
SPL (cell F2): =D2+E2
RPL (cell G2): =PERCENTRANK.INC($F$2:$F$5, F2, 15)R
d <- data.frame(x = c(10, 25, 40, 18), y = c(5, 3, 20, 12), row.names = LETTERS[1:4])
pr <- function(v) (rank(v, ties.method = "min") - 1) / (length(v) - 1)
d$SPL <- pr(d$x) + pr(d$y)
d$RPL <- pr(d$SPL)
dPython (pandas)
import pandas as pd
d = pd.DataFrame({"x": [10, 25, 40, 18], "y": [5, 3, 20, 12]}, index=list("ABCD"))
pr = lambda s: (s.rank(method="min") - 1) / (len(s) - 1)
d["SPL"] = pr(d["x"]) + pr(d["y"])
d["RPL"] = pr(d["SPL"])
print(d)FAQ
What is the CDC Social Vulnerability Index?
It is a CDC/ATSDR ranking of US census tracts by relative social vulnerability, built from 16 ACS variables in four themes and expressed as percentile rankings from 0 to 1.
How do I read a Social Vulnerability Index map?
Darker or higher classes mean a higher percentile ranking, that is, greater relative vulnerability compared with other tracts in the chosen database. Check whether the map uses the national or a state database and which release year.
Is the SVI a probability of harm?
No. It is a relative ranking. It does not predict the harm a given hazard will cause.
Which variables does the SVI use?
Sixteen: five socioeconomic, five household characteristics, one racial and ethnic minority status, and five housing type and transportation variables (see the table above).
Where can I get SVI data for my area?
CDC/ATSDR publish the SVI databases and documentation on their Place and Health pages. For public health context, see what is epidemiology and the attack rate term.
Sources
- CDC/ATSDR. CDC/ATSDR Social Vulnerability Index 2022 documentation. https://www.atsdr.cdc.gov/place-health/media/pdfs/2024/10/SVI2022Documentation.pdf (read 2026-11-07)
- CDC/ATSDR. SVI data and documentation download page. https://www.atsdr.cdc.gov/place-health/php/svi/svi-data-documentation-download.html (read 2026-11-07)
- Flanagan, B. E., et al. (2011). A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management, 8(1). https://doi.org/10.2202/1547-7355.1792
